[BUG] Retry workflow from task inside fork

**Describe the bug**
Inside the fork when you retry the workflow from a step, the separate branches will be deleted too but will not run later.
As the attached image, every part of the fork has built the same way, a step followed by a sub workflow. After the initial execution the Sequence numbers will be as shown in the picture (the steps will be 1, 2, 3 and sub workflows 4, 5, 6) which results that the rerun logic will filter out everything after the retried step (Seq 2). This is totally invalid as inside the fork the "next" branches has nothing to do with the original step (_Step 2_ in this case)

**Details**
Conductor version: 3.21.5

**To Reproduce**
I cannot share the exact workflow because it contains confidental information but attached a screenshot which clearly describes the schema of the workflow.
After running the workflow you should retry the workflow from the task _Step 2_ which I have done through the API.

**Expected behavior**
Only the selected branch should be restarted and the others should not be reseted. In this case _Step 2_ and _Sub Workflow 2_

**Screenshots**

Start by reproducing the API retry-from-task scenario with a forked workflow matching the attached schema, since the exact workflow is unavailable. Trace how retrying Step 2 filters subsequent sequence numbers; done means only Step 2 and Sub Workflow 2 restart while sibling branches remain intact.

Written by the indexing model from the issue text.